The ~20 most important skills for Luci's daily operations, grouped by category. For the full list of 110 active skills, run
ls ~/.claude/skills/.
| Skill | Description | Triggers |
|---|---|---|
| dev-loop | Universal entry point for ALL code changes. Classifies into Tier 1 (one-shot), Tier 2 (plan+build+review), or Tier 3 (research+TDD+council). Mandatory for every ticket worker. | Any MC ticket pickup, "dev loop", "plan and build", "TDD" |
| investigate | Systematic debugging with root cause analysis. Four phases: investigate, analyze, hypothesize, implement. Iron law: no fixes without root cause. | "debug this", "fix this bug", "why is broken", "root cause" |
| review | Pre-landing PR review. Checks for SQL safety, LLM trust boundary violations, conditional side effects. | "review this PR", "code review", "pre-landing review" |
| qa | Systematic QA testing of web apps. Finds bugs, then iteratively fixes them with atomic commits and re-verification. | "qa", "test this site", "find bugs", "test and fix" |
| ship | Ship workflow: merge base branch, run tests, review diff, bump VERSION, update CHANGELOG, commit, push, create PR. | "ship", "deploy", "push to main", "create a PR" |
| playwright | Browser automation and E2E testing with Playwright CLI. Auto-detects dev servers, writes test scripts. | "playwright", "e2e test", "browser test", "test login flow" |
| Skill | Description | Triggers |
|---|---|---|
| deep-research | Multi-source research with synthesis. Handles comparisons, competitive analysis, best practices, community sentiment. | "research", "investigate", "compare options", "what's the best way to", "X vs Y" |
| second-opinion | Council of AI models (Gemini + Codex minimum, all 4 for architecture). Single model for debugging, full council for design reviews. | "second opinion", "ask codex", "ask gemini", "council review", "stuck on this bug" |
| munger | Run decisions through Charlie Munger's mental models -- 25 cognitive biases, inversion, latticework analysis. | Evaluating decisions, investments, strategies, partnerships |
| Skill | Description | Triggers |
|---|---|---|
| xlsx | Read, edit, create spreadsheets (.xlsx, .xlsm, .csv, .tsv). Handles formulas, formatting, charting, data cleaning. | Any spreadsheet file reference, "the xlsx", "update the spreadsheet" |
| Full PDF toolkit: read, merge, split, rotate, watermark, fill forms, encrypt/decrypt, OCR scanned documents. | Any .pdf file reference, "combine PDFs", "extract from PDF" | |
| docx | Create, read, edit Word documents with professional formatting. | "Word doc", ".docx", "create a document" |
| pptx | Create and manipulate PowerPoint presentations. | Any .pptx file reference, "slide deck", "presentation" |
| gemini-transcribe | Transcribe audio files using Google Gemini with speaker diarization and multi-language support. | "transcribe", "speech-to-text", "meeting transcription" |
| Skill | Description | Triggers |
|---|---|---|
| email-index | Query the indexed email database (2,026 emails across 23 projects) on Luci via SSH. Historical lookups only. | "emails about", "emails from", "email history", "past emails" |
| inbox-assist | Interactive email assistant for Gmail and MS365 Outlook. Scans inbox, shows actionable emails, does full work (attachments, research, drafting). | "check email", "inbox", "email assistant" |
| m365 | Microsoft 365 integration: Outlook email, calendar, Teams, OneDrive, tasks. | "send email", "check calendar", "meeting availability" |
| Skill | Description | Triggers |
|---|---|---|
| browse | Fast headless browser (~100ms/command). Navigate URLs, interact with elements, take annotated screenshots, test responsive layouts, verify deployments. | "open in browser", "test the site", "check the page", "screenshot" |
| defuddle | Extract clean markdown from web pages using Defuddle CLI. Removes clutter and navigation to save tokens. | Any URL to read or analyze, documentation pages, articles |
| Skill | Description | Triggers |
|---|---|---|
| luci | Dispatch work to Luci. Schedule recurring tasks, check status, retrieve results from completed tasks. | "luci", "background", "schedule", "overnight", "run this later" |
| self-improve | Scans session transcripts for unsaved learnings, corrections, friction, and repeated patterns. Proposes saves to memory, rules, skills, CLAUDE.md. | "self-improve", "review sessions", "what did we learn" |
| health | Code quality dashboard. Runs type checker, linter, test runner, dead code detector. Computes weighted 0-10 score with trend tracking. | "health check", "code quality", "how healthy" |
| Skill | Description | Triggers |
|---|---|---|
| safair-dashboard | Build the FlySafair CEO HTML dashboard with charts, KPIs, and interactive reports. | "CEO dashboard", "update dashboard", "rebuild dashboard" |
| encft-finance | Financial management for Portfolio.xlsx, entity bookkeeping (Werda, ENC Family Trust), group financials. | "portfolio", "financial statements", "entity books" |
| booking-curve | Pull FlySafair booking curve data from Neon PostgreSQL. | "booking curve", "pull booking data" |
Mission Control is the board for your delegated work: requests come in, Luci coordinates the next step, and evidence stays visible for review.
Luci is your always-on assistant for routing, status updates, and follow-through. Operators can still open deeper evidence when needed.